Are Some Teachers AIs?

News that a computer program did better than most humans at questions on IQ tests has been going around the blogosphere. As far as I could tell, the program worked by examining published texts to see which words usually went with which. Let's consider what happens when humans use such an algorithm.

If we search high and low for examples of antonyms in published texts, we frequently see that pairs of words identified as antonyms occur in the form of “Word_A and Word_B.” That's as plain as black and white. If we generalize from that we might decide, for example, that “cat” is the opposite of “dog” or “tadpole” is the opposite of “frog.” I think the algorithm has limitations.

ObSF: “Camels and Dromedaries, Clem” by R. A. Lafferty